[ENG-3771] projects-cli-0: explicit projects CLI

README.md

@@ -155,6 +155,10 @@ Prime Lab connects verifiers environments to evaluations, GEPA prompt optimizati
 # Set up a Lab workspace
 prime lab setup
 
+# Create a Lab project and make it active in this workspace
+prime project create "Alphabet Sort Baselines"
+prime project current
+
 # List trainable models, capacity, and token pricing
 prime train models
 
@@ -163,6 +167,7 @@ prime train init
 
 # Launch the run from the generated config
 prime train rl.toml
+prime train rl.toml --project <project-id>
 
 # Inspect and manage Hosted Training runs
 prime train list
@@ -171,6 +176,26 @@ prime train metrics <run-id>
 prime train checkpoints <run-id>
 ```
 
+Lab projects group related training runs, evaluations, and adapters. Use
+`prime project use <project-id>` to switch the active workspace project, or
+`prime project clear` to clear it. Existing runs and adapters support project
+add/remove/clear; evaluations support assign/clear.
+
+```bash
+# Manage projects
+prime project list
+prime project show <project-id>
+prime project update <project-id> --description "Baseline alphabet sort runs"
+
+# Attach existing artifacts
+prime project assign run <run-id> <project-id>
+prime project remove run <run-id> <project-id>
+prime project assign adapter <adapter-id> <project-id>
+prime project remove adapter <adapter-id>  # clear all adapter project memberships
+prime project assign eval <eval-id> <project-id>
+prime project remove eval <eval-id>        # clear the evaluation project
+```
+
 ### GPU Resources
 
 ```bash
@@ -211,6 +236,7 @@ prime eval push
 
 # Push specific eval directory (verifiers format)
 prime eval push outputs/evals/gsm8k--gpt-4/abc123
+prime eval push outputs/evals/gsm8k--gpt-4/abc123 --project <project-id>
 
 # Push a public evaluation (default is private)
 prime eval push --public

packages/prime-evals/README.md

@@ -37,6 +37,7 @@ eval_response = client.create_evaluation(
     model_name="gpt-4o-mini",
     dataset="gsm8k",
     framework="verifiers",
+    project_id="project-id",
     metadata={
         "version": "1.0",
         "num_examples": 10,
@@ -220,6 +221,29 @@ client.finalize_evaluation(eval_id, metrics=eval_data.get("metrics"))
 print(f"Successfully pushed evaluation: {eval_id}")
 ```
 
+## Project Attachment
+
+Evaluations can be created inside a Lab project, moved to another project, or
+cleared from their project. Evaluation assignment is set/clear; targeted removal
+from one project is not supported for evaluations.
+
+```python
+eval_response = client.create_evaluation(
+    name="gsm8k-project-baseline",
+    environments=[{"id": "gsm8k"}],
+    model_name="gpt-4o-mini",
+    project_id="project-id",
+)
+
+eval_id = eval_response["evaluation_id"]
+
+# Move the evaluation to another project
+client.update_evaluation(eval_id, project_id="another-project-id")
+
+# Clear the evaluation project
+client.update_evaluation(eval_id, clear_project=True)
+```
+
 ## API Reference
 
 ### EvalsClient
@@ -232,6 +256,7 @@ Main client for interacting with the Prime Evals API.
 - `push_samples()` - Push evaluation samples
 - `finalize_evaluation()` - Finalize an evaluation with final metrics
 - `get_evaluation()` - Get evaluation details by ID
+- `update_evaluation()` - Update evaluation details or assign/clear a project
 - `list_evaluations()` - List evaluations with optional filters
 - `get_samples()` - Get samples for an evaluation
 
@@ -276,4 +301,3 @@ except EvalsAPIError as e:
 ## License
 
 MIT License - see LICENSE file for details
-

packages/prime-evals/src/prime_evals/evals.py

@@ -74,9 +74,9 @@ def _lookup_environment_by_slug(self, owner_slug: str, name: str) -> str:
             EvalsAPIError: If the environment does not exist (404)
         """
         try:
-            response = self.client.get(f"/environmentshub/{owner_slug}/{name}/@latest")
-            details = response.get("data", response)
-            return details["id"]
+            lookup_data: Dict[str, Any] = {"name": name, "owner_slug": owner_slug}
+            response = self.client.post("/environmentshub/lookup", json=lookup_data)
+            return response["data"]["id"]
         except APIError as e:
             raise EvalsAPIError(
                 f"Environment '{owner_slug}/{name}' does not exist in the hub. "
@@ -159,6 +159,7 @@ def create_evaluation(
         task_type: Optional[str] = None,
         description: Optional[str] = None,
         tags: Optional[List[str]] = None,
+        project_id: Optional[str] = None,
         metadata: Optional[Dict[str, Any]] = None,
         metrics: Optional[Dict[str, Any]] = None,
         is_public: Optional[bool] = None,
@@ -202,6 +203,7 @@ def create_evaluation(
             "task_type": task_type,
             "description": description,
             "tags": tags or [],
+            "project_id": project_id,
             "metadata": metadata,
             "metrics": metrics,
         }
@@ -365,6 +367,8 @@ def update_evaluation(
         task_type: Optional[str] = None,
         description: Optional[str] = None,
         tags: Optional[List[str]] = None,
+        project_id: Optional[str] = None,
+        clear_project: bool = False,
         metadata: Optional[Dict[str, Any]] = None,
         metrics: Optional[Dict[str, Any]] = None,
     ) -> Dict[str, Any]:
@@ -375,11 +379,16 @@ def update_evaluation(
             "framework": framework,
             "task_type": task_type,
             "description": description,
-            "tags": tags if tags is not None else [],
+            "tags": tags,
+            "project_id": project_id,
             "metadata": metadata,
             "metrics": metrics,
         }
-        payload = {k: v for k, v in payload.items() if v is not None or k in ["tags"]}
+        payload = {
+            k: v
+            for k, v in payload.items()
+            if v is not None or (clear_project and k == "project_id")
+        }
 
         response = self.client.request("PUT", f"/evaluations/{evaluation_id}", json=payload)
         return response
@@ -424,9 +433,9 @@ async def _lookup_environment_by_slug(self, owner_slug: str, name: str) -> str:
             EvalsAPIError: If the environment does not exist (404)
         """
         try:
-            response = await self.client.get(f"/environmentshub/{owner_slug}/{name}/@latest")
-            details = response.get("data", response)
-            return details["id"]
+            lookup_data: Dict[str, Any] = {"name": name, "owner_slug": owner_slug}
+            response = await self.client.post("/environmentshub/lookup", json=lookup_data)
+            return response["data"]["id"]
         except APIError as e:
             raise EvalsAPIError(
                 f"Environment '{owner_slug}/{name}' does not exist in the hub. "
@@ -515,6 +524,7 @@ async def create_evaluation(
         task_type: Optional[str] = None,
         description: Optional[str] = None,
         tags: Optional[List[str]] = None,
+        project_id: Optional[str] = None,
         metadata: Optional[Dict[str, Any]] = None,
         metrics: Optional[Dict[str, Any]] = None,
         is_public: Optional[bool] = None,
@@ -558,6 +568,7 @@ async def create_evaluation(
             "task_type": task_type,
             "description": description,
             "tags": tags or [],
+            "project_id": project_id,
             "metadata": metadata,
             "metrics": metrics,
         }
@@ -715,6 +726,8 @@ async def update_evaluation(
         task_type: Optional[str] = None,
         description: Optional[str] = None,
         tags: Optional[List[str]] = None,
+        project_id: Optional[str] = None,
+        clear_project: bool = False,
         metadata: Optional[Dict[str, Any]] = None,
         metrics: Optional[Dict[str, Any]] = None,
     ) -> Dict[str, Any]:
@@ -725,11 +738,16 @@ async def update_evaluation(
             "framework": framework,
             "task_type": task_type,
             "description": description,
-            "tags": tags if tags is not None else [],
+            "tags": tags,
+            "project_id": project_id,
             "metadata": metadata,
             "metrics": metrics,
         }
-        payload = {k: v for k, v in payload.items() if v is not None or k in ["tags"]}
+        payload = {
+            k: v
+            for k, v in payload.items()
+            if v is not None or (clear_project and k == "project_id")
+        }
 
         response = await self.client.request("PUT", f"/evaluations/{evaluation_id}", json=payload)
         return response

packages/prime-evals/src/prime_evals/models.py

@@ -32,6 +32,7 @@ class Evaluation(BaseModel):
     run_id: Optional[str] = Field(None, alias="runId")
     version_id: Optional[str] = Field(None, alias="versionId")
     tags: List[str] = Field(default_factory=list)
+    project_id: Optional[str] = Field(None, alias="projectId")
     metadata: Optional[Dict[str, Any]] = None
     metrics: Optional[Dict[str, Any]] = None
     total_samples: Optional[int] = Field(None, alias="totalSamples")
@@ -66,6 +67,7 @@ class CreateEvaluationRequest(BaseModel):
     task_type: Optional[str] = None
     description: Optional[str] = None
     tags: List[str] = Field(default_factory=list)
+    project_id: Optional[str] = None
     metadata: Optional[Dict[str, Any]] = None
     metrics: Optional[Dict[str, Any]] = None
 

packages/prime-evals/tests/test_evals.py

@@ -125,45 +125,6 @@ def test_sample_model_with_metadata():
     assert sample.info == {"batch": 1}
 
 
-def test_lookup_environment_by_slug_uses_owner_aware_detail_endpoint():
-    calls = []
-
-    class FakeAPIClient:
-        def get(self, endpoint):
-            calls.append(("get", endpoint))
-            return {"data": {"id": "env-123"}}
-
-        def post(self, *_args, **_kwargs):
-            raise AssertionError("owner/name lookup must not use team_slug lookup")
-
-    client = EvalsClient(FakeAPIClient())
-
-    env_id = client._lookup_environment_by_slug("d42me", "opencode-cp")
-
-    assert env_id == "env-123"
-    assert calls == [("get", "/environmentshub/d42me/opencode-cp/@latest")]
-
-
-def test_async_lookup_environment_by_slug_uses_owner_aware_detail_endpoint():
-    calls = []
-
-    class FakeAsyncAPIClient:
-        async def get(self, endpoint):
-            calls.append(("get", endpoint))
-            return {"data": {"id": "env-123"}}
-
-        async def post(self, *_args, **_kwargs):
-            raise AssertionError("owner/name lookup must not use team_slug lookup")
-
-    client = AsyncEvalsClient.__new__(AsyncEvalsClient)
-    client.client = FakeAsyncAPIClient()
-
-    env_id = asyncio.run(client._lookup_environment_by_slug("d42me", "opencode-cp"))
-
-    assert env_id == "env-123"
-    assert calls == [("get", "/environmentshub/d42me/opencode-cp/@latest")]
-
-
 def test_push_samples_reports_progress_and_reuses_http_client(monkeypatch):
     posts = []
     created_clients = []
@@ -271,6 +232,101 @@ def test_evals_client_context_manager():
         pass  # Expected to fail without proper initialization
 
 
+def test_lookup_environment_by_slug_uses_owner_slug_payload():
+    captured = {}
+
+    class DummyHTTPClient:
+        def post(self, endpoint, json=None):
+            captured["endpoint"] = endpoint
+            captured["json"] = json
+            return {"data": {"id": "env-123"}}
+
+    client = EvalsClient.__new__(EvalsClient)
+    client.client = DummyHTTPClient()
+
+    assert client._lookup_environment_by_slug("alice", "gsm8k") == "env-123"
+    assert captured == {
+        "endpoint": "/environmentshub/lookup",
+        "json": {"name": "gsm8k", "owner_slug": "alice"},
+    }
+
+
+def test_async_lookup_environment_by_slug_uses_owner_slug_payload():
+    captured = {}
+
+    class DummyHTTPClient:
+        async def post(self, endpoint, json=None):
+            captured["endpoint"] = endpoint
+            captured["json"] = json
+            return {"data": {"id": "env-123"}}
+
+    client = AsyncEvalsClient.__new__(AsyncEvalsClient)
+    client.client = DummyHTTPClient()
+
+    result = asyncio.run(client._lookup_environment_by_slug("alice", "gsm8k"))
+
+    assert result == "env-123"
+    assert captured == {
+        "endpoint": "/environmentshub/lookup",
+        "json": {"name": "gsm8k", "owner_slug": "alice"},
+    }
+
+
+def test_create_evaluation_sends_project_id_payload():
+    captured = {}
+
+    class DummyConfig:
+        team_id = None
+
+    class DummyHTTPClient:
+        config = DummyConfig()
+
+        def request(self, method, endpoint, json=None, params=None):
+            captured["method"] = method
+            captured["endpoint"] = endpoint
+            captured["json"] = json
+            captured["params"] = params
+            return {"evaluation_id": "eval-123"}
+
+    client = EvalsClient.__new__(EvalsClient)
+    client.client = DummyHTTPClient()
+
+    response = client.create_evaluation(
+        name="gsm8k",
+        run_id="run-123",
+        model_name="gpt-4o-mini",
+        project_id="project-123",
+    )
+
+    assert response == {"evaluation_id": "eval-123"}
+    assert captured["method"] == "POST"
+    assert captured["endpoint"] == "/evaluations/"
+    assert captured["json"]["project_id"] == "project-123"
+    assert "projectId" not in captured["json"]
+
+
+def test_update_evaluation_clear_project_sends_null_project_id():
+    captured = {}
+
+    class DummyHTTPClient:
+        def request(self, method, endpoint, json=None, params=None):
+            captured["method"] = method
+            captured["endpoint"] = endpoint
+            captured["json"] = json
+            captured["params"] = params
+            return {"evaluation_id": "eval-123"}
+
+    client = EvalsClient.__new__(EvalsClient)
+    client.client = DummyHTTPClient()
+
+    response = client.update_evaluation("eval-123", clear_project=True)
+
+    assert response == {"evaluation_id": "eval-123"}
+    assert captured["method"] == "PUT"
+    assert captured["endpoint"] == "/evaluations/eval-123"
+    assert captured["json"] == {"project_id": None}
+
+
 def test_evaluation_model_minimal():
     """Test Evaluation model with minimal data"""
     data = {